• Keine Ergebnisse gefunden

The NICE Cost-Effectiveness Threshold: What it is and What that Means

N/A
N/A
Protected

Academic year: 2022

Aktie "The NICE Cost-Effectiveness Threshold: What it is and What that Means"

Copied!
13
0
0

Wird geladen.... (Jetzt Volltext ansehen)

Volltext

(1)

The NICE Cost-Effectiveness Threshold:

What it is and What that Means

McCabe, C and Claxton, K and Culyer, AJ

univereity of Leeds, university of York

2008

Online at https://mpra.ub.uni-muenchen.de/26466/

MPRA Paper No. 26466, posted 05 Nov 2010 20:54 UTC

(2)

The NICE

Cost-Effectiveness Threshold

What it is and What that Means

Christopher McCabe,1 Karl Claxton2 and Anthony J. Culyer3,4

1 Academic Unit of Health Economics and NICE Decision Support Unit, Leeds Institute of Health Sciences, University of Leeds, Leeds, UK

2 Centre for Health Economics, Department of Economics and NICE Decision Support Unit, University of York, Heslington, York, UK

3 Centre for Health Economics, Department of Economics, University of York, Heslington, York, UK

4 Department of Health Policy Management and Evaluation, University of Toronto, Toronto, Ontario, Canada

The National Institute for Health and Clinical Excellence (NICE) has been

Abstract

using a cost-effectiveness threshold range between £20 000 and £30 000 for over 7 years. What the cost-effectiveness threshold represents, what the appropriate level is for NICE to use, and what the other factors are that NICE should consider have all been the subject of much discussion. In this article, we briefly review these questions, provide a critical assessment of NICE’s utilization of the incre- mental cost-effectiveness ratio (ICER) threshold to inform its guidance, and suggest ways in which NICE’s utilization of the ICER threshold could be developed to promote the efficient use of health service resources.

We conclude that it is feasible and probably desirable to operate an explicit single threshold rather than the current range; the threshold should be seen as a threshold at which ‘other’ criteria beyond the ICER itself are taken into account;

interventions with a large budgetary impact may need to be subject to a lower threshold as they are likely to displace more than the marginal activities; reim- bursement at the threshold transfers the full value of an innovation to the manufacturer.

Positive decisions above the threshold on the grounds of innovation reduce population health; the value of the threshold should be reconsidered regularly to ensure that it captures the impact of changes in efficiency and budget over time;

the use of equity weights to sustain a positive recommendation when the ICER is above the threshold requires knowledge of the equity characteristics of those patients who bear the opportunity cost. Given the barriers to obtaining this knowledge and knowledge about the characteristics of typical beneficiaries of UK NHS care, caution is warranted before accepting claims from special pleaders;

uncertainty in the evidence base should not be used to justify a positive recom- mendation when the ICER is above the threshold. The development of a pro- gramme of disinvestment guidance would enable NICE and the NHS to be more confident that the net health benefit of the Technology Appraisal Programme is positive.

(3)

1. Background automatically be defined as not cost effective or below which it would. Given the fixed budget of the The National Institute for Health and Clinical NHS, the appropriate threshold is that of the oppor- Excellence (NICE) is charged with the task of con- tunity cost of programmes displaced by new, more sidering both the effectiveness and cost effective- costly technologies. However, estimating this thres- ness of treatments and then making recommenda- hold would require complete information about the tions as to the provision of such treatments within costs and QALYs from all competing healthcare the UK NHS. Cost-effectiveness analysis assesses programmes and the Committee does not have this two or more alternative courses of action in terms of information. Furthermore, the threshold will change their costs and benefits. The comparison is summa- over time as the budget for healthcare changes.

rized using the expected incremental cost-effective- Although the use of a threshold is inappropriate, ness ratio (ICER). This is a measure of the addition- comparisons of the most plausible ICER of a partic- al cost per additional unit of health gain produced by ular technology compared with other programmes one intervention compared with another. NICE’s that are currently funded are possible and are a preferred form of cost-effectiveness analysis uses legitimate reference for the Committee.”[1]

the QALY to describe the outcome of each interven- This statement acknowledges the importance of tion. By extension, the preferred form of ICER is the considering the opportunity cost of implementing cost per QALY gained. Within the NICE appraisal new treatments given a fixed threshold – whilst process, the ICER for each technology is compared conversely suggesting that, since the data required with a threshold value (generally accepted as having to estimate the threshold quantitatively are not avail- an upper limit of £30 000) to establish whether the able, it is inappropriate to use a threshold. Our technology represents an efficient use of limited interpretation of this apparently contradictory state- NHS resources. ment is that it is use of a particular threshold that is The objective of this article is to review the to be avoided, hence NICE’s emphasis on a range.

current state of knowledge regarding the cost-effec- The guide then goes on to consider a range of tiveness threshold, the principles of its use in health- possible other factors to take into account in cases of care resource allocation decisions and any argu- technologies with ICERs at the lower and upper ments for and against changing the threshold from boundary of the range:

the current range of £20 000–30 000. “Below a most plausible ICER of £20,000/QALY, Section 2 summarizes the statements in the 2004 judgements about the acceptability of a technology Guide to the Methods of Technology Appraisal re- as an effective use of NHS resources are based garding the value and use of the cost-effectiveness primarily on the cost-effectiveness estimate. Above a threshold. Section 3 reviews the relevant literature most plausible ICER of £20,000/QALY, judgements on the use of the cost-effectiveness threshold in about the acceptability of the technology as an effec- resource allocation decision making. Section 4 dis- tive use of NHS resources are more likely to make cusses key implications of using an ICER threshold more explicit reference to factors including:

to promote population health gain from the NHS

the degree of uncertainty surrounding the calcu- budget. Section 5 considers the issue of whether the lation of ICERs

NICE threshold should change; and section 6 at-

the innovative nature of the technology tempts to summarize the key observations of the

the particular features of the condition and popu- paper.

lation receiving the technology

where appropriate, the wider societal costs and 2. What the Current Methods Guide Says

benefits.”[1]

The 2004 Methods Guide[1] refers several times This approach echoes ideas advanced by to the cost-effectiveness threshold. In chapter 6 Akehurst[2] in 2002, and seems to imply that NICE’s (page 33), it states: ‘effective threshold’ is actually £20 000 per QALY.

“The Appraisal Committee does not use a fixed When cost-effectiveness ratios for a treatment ex- ICER threshold above which a technology would ceed this, the Appraisal Committee considers

(4)

(i) whether the characteristics of the condition or whilst the latter treated NICE decisions as discrete population receiving the treatment would lead them choice experiments.

to value the health gain produced by the intervention Rawlins and Culyer[5] identified an increasing more highly than the estimate made in the analysis; likelihood of rejection as the ICER increased be- (ii) whether innovative characteristics of the inter- yond £15 000, with few interventions being ap- vention are such as to require explicit consideration proved with an ICER >£30 000 (figure 1). Devlin of the Secretary of State’s instruction to give due and Parkin,[6] in contrast, estimated the threshold to weight to innovativeness, despite the excess oppor- be “somewhat higher than the £20,000 – £30,000 tunity cost from a purely efficiency perspective; and which NICE has publicly identified.”

(iii) whether other benefits to society, outside of There are several problems with basing the cur- those considered by the cost-effectiveness analysis, rent threshold on previous decisions. First, it is not are such that it is ‘socially desirable’ for the treat- necessarily desirable for current decisions to use the ment to be made available. same decision rule as for previous ones; consistency The proposed role of uncertainty in decision of decision rule (the cost-effectiveness threshold) making is unclear. The text may mean that when the can conflict with consistency of objective (maximiz- ICER exceeds the lower bound of the threshold ing expected health gain). We discuss in more detail range, the committee will seek greater levels of why the threshold might change over time in section certainty to support a positive recommendation. 4. Second, this approach requires that previous deci- However, the text is also consistent with Akehurst’s sions either took no account of any ‘other’ consider- proposal[3] that when there is great uncertainty about ations or that any such consideration was judged not an ICER in excess of the threshold value, it is sufficient to have an impact on the decision, other- appropriate to treat the estimate as not significantly wise the linking of particular ICERs to a particular different from the threshold value. threshold value will have (largely unknowable) bi-

Use of additional criteria is not inconsistent with ases.

the operation of an explicit, single threshold value,

3.2 Setting to Determine the Optimal nor is it inconsistent with much of the literature on

Healthcare Budget social preferences over healthcare resource alloca-

tion.[3] However, as we discuss in section 4.2, there

Some have suggested that the appropriate process are substantive issues concerning the ways in which

is to identify the marginal value that society attaches such additional considerations should be operation-

to health.[7] NICE itself has promoted two research alized.

projects to examine what value people in the UK attach to an additional QALY. If the cost-effective- 3. Setting the Threshold

ness threshold were set by such an empirically re- There is significant argument about how the cost-

effectiveness threshold should be determined. Three broad approaches have been proposed: (i) it should be inferred from previous decisions; (ii) it should be set so as to determine the optimal healthcare budget;

and (iii) it should be set so as to exhaust an exoge- nously determined budget.[4]

3.1 Inferring the Value from Previous Decisions

Rawlins and Culyer[5] and Devlin and Parkin[6]

made two attempts to infer NICE’s cost-effective- ness threshold from reviews of previous decisions.

The former was an essentially qualitative analysis,

Increasing cost per QALY (log scale)

Probability of rejection

Fig. 1. Relationship between cost effectiveness and probability of rejection (reproduced from Rawlins and Culyer[5]).

(5)

vealed monetary value, the implication is that inter- sistent with the value attached to a life in other parts ventions having an ICER below that value should be of the public sector. Whilst health and life are the approved. The healthcare budget would then be primary (although not sole) objectives of the NHS, whatever sum was required to implement the they are not the primary objectives of other public purchases. Setting the threshold would thus effec- sector activities. The budgets allocated to these dif- tively determine the NHS’s budget.[8] The budget ferent activities by parliament imply a relative valu- would be demonstrably consistent with the value ation of these objectives as well as the impact on that ‘society’ attached to health, and the state would health and length of life. It would be a major task to be committed to increase the budget so long as the isolate the ‘health component’ in these other activi- ICERs for new interventions fell below the thres- ties. Currently, only transport uses an explicit value

hold. and, as one of the authors of this briefing paper has

Three approaches have been suggested for quan- previously observed, “NICE simply does not have tifying the marginal value of health: (i) discovering (and nor is it mandated to acquire) the kind of the willingness to pay for health gain of a represen- information about outputs in non-health sectors that tative sample of society;[7] (ii) using the value of life/ it would need to form necessary judgements about health employed in other areas of public sector the marginal costs and benefits of health spending resource allocation;[9] and (iii) setting it equal to versus spending in other areas of public services.”[4]

GDP per capita.[10] Thus, although it may be intuitively appealing, it is If there was a direct link between society’s will- not feasible for the threshold to be set (by NICE) by ingness to pay for health gain and the budget of the reference to other public sector activities.

healthcare system, setting the threshold with refer-

Williams[10] suggested that a ‘common sense’

ence to it would seem appropriate. However, in the

value for the threshold would be per capita GDP. At UK, as in many other countries, the budget of the

the time of lecture, this was somewhat lower than healthcare system is determined in large part by

the bottom of the threshold range used by NICE.

parliament and is done (doubtless imperfectly) by

The appeal of this proposal is that, if every member broad assessments of the marginal value of exten-

of society were to be given a ‘fair share’ of a sions of a wide variety of public programmes and of

nation’s wealth, they would receive the per capita the value of purchasing power left in the pockets of

GDP. The maximum they could therefore spend on consumers. The budget allocated to healthcare by

health gain in any one year would be the per capita parliament, therefore, already contains an implicit

GDP. Three significant problems present them- value of marginal health gain (MHG) – relative to

selves. First, the approach implies that the society alternative uses of public funds. It is difficult to see

might be willing to devote all its wealth to health- how experimental methods for revealing the social

care, which is manifestly not the case. Second, the value of a QALY could capture these opportunity

same thought experiment will yield the same ‘maxi- costs more effectively (or more legitimately) than

mum’ for any good or service in GDP and so pro- parliament. To substitute the ‘direct democracy’ of

vides no basis whatever for choosing between any of public opinion for a parliamentary process plainly

them. Third, the average cost effectiveness of also raises constitutional issues well beyond the

healthcare can be at or below per capita GDP with scope of this article.

the cost effectiveness of marginal programmes be- The healthcare system is not the only area of

ing markedly higher thanks to diminishing marginal public policy concerned with promoting health. For

returns. NICE has to establish whether a new inter- example, transport investment decisions typically

vention is more cost effective than the marginal take account of the expected impact on injury and

interventions that would have to be displaced in death rates when appraising road building schemes.

order to pay for it from a constant budget. Use of an It is therefore intuitively appealing that the value of

arbitrary average risks rejecting interventions that health ought to be consistent across public sector

were more cost effective than those already pro- activities. Loomes[9] has suggested that the cost-

vided.

effectiveness threshold should be set at a level con-

(6)

3.3 Setting to Optimally Exhaust a Budget available technologies: those in use as well as those that could be used but are not, ordered by contribu- From the beginning, NICE’s use of cost-effec- tion to health gain.

tiveness analysis has been perceived as a means of It is immediately apparent that NICE confronts promoting the efficient use of available NHS resour- three potentially interesting MHGs, the size of none

ces.[11,12] The cost-effectiveness threshold ought of which it can be sure. E–a is the actual MHG

thus to be the cost per QALY of the least efficient implied by current use in the NHS. It is what the funded treatment (i.e. the intervention with the high- current ‘threshold’ would appear to be if a compre- est cost per QALY). For a new intervention to add to hensive assessment were to be made of the ways in health, it must be more efficient per unit of resource which current NHS resources are used. E–c is the than the least efficient currently funded intervention health gain to be achieved from adopting the best and ought to displace it, in whole or part, so that the technology not currently in use in the NHS. E–b is marginal productivity of each intervention in terms the threshold above which technologies ought to be of health was everywhere equalized. Here too, how- adopted and below which they ought not. The incor- ever, is another evident informational challenge. If poration of any technology not in current use with an identifying the marginal interventions for disinvest- MHG above E–b would represent an increase in ment is too difficult, the threshold requires an alter- health outcomes as long as it displaces a technology native justification.[12] Here, we need to tread with with a lower MHG (in the range E’–E). The optimal care. On the one hand, there is an issue of principle: solution is plainly to cease using all those technolo- what the threshold ought to represent, a value judge- gies in the range E’–E on H–a and substitute for ment; and on the other, an empirical question: the them all those in the range E–E" (= E’–E) on c–f.

value it should take in any specific context. If the function of NICE is to substitute more In figure 2, the range of technologies in O–E efficient interventions for less efficient ones, it can embodies those extant and provided in the NHS. Let do this through specifying a ‘working’ cost-effec- us assume that all provide positive MHG and that tiveness threshold, reflecting the Institute’s estimate the least productive one has an MHG of E–a. How- of the ICER of the least cost-effective activity ever, there are many technologies either extant or undertaken by the NHS. This working estimate can emergent that are not currently provided within the be drawn from (i) the incomplete evidence base on NHS. These technologies are ranked in a separate the cost effectiveness of interventions that the NHS downward-sloping function to the right of E, la- does provide; and (ii) stakeholders’ personal and belled c–f. A composite MHG curve is the horizon- professional knowledge of the likely value of funded tal sum of the two lines, H–d–e, which combines all interventions, for which formal evaluations are not available.[1,8] Over time, this ‘working’ ICER can be adjusted in a casuistical fashion reflecting develop- ments in the published evidence base and evidence on the efficiency of disinvestments made to fund the recommended interventions, and changes in the healthcare budget and in judgments about the effi- ciency of healthcare production.

Although the ‘threshold-searcher’ model of Culyer et al.[8] describes how resource allocation processes can utilize ICERs for healthcare resource allocation decisions at the margin, the authors did not address the frequently cited criticisms of Birch and Gafni,[13-16] who have repeatedly argued that decision makers cannot maximize health gain from limited resources by using ICERs in isolation from information on budget impact. To do so, they say, is

O

Marginal health gain

UK NHS expenditures per year E

a b d c

H

e

E’ E” f

Fig. 2. The National Institute for Health and Clinical Excellence as a threshold searcher (reproduced from Culyer et al.,[8] with permis- sion).

(7)

a recipe for “continued expansion of expendi- technologies will have to be displaced). In this way, ture.”[13-16] Their argument is that the opportunity allocation processes based on a cost-effectiveness cost of a positive decision is determined by the total threshold can fully capture the opportunity cost of budgetary impact, not the ICER. It is possible for the both positive and negative investment recommenda- total budget for an existing health intervention to be tions.

less than that for the new intervention even though The rate of change in the productivity of health- the new intervention has a lower ICER. If the new care also matters. Thus, when productivity is rising intervention is mandated on the basis of the ICER through the use of relatively efficient technologies, alone, then extra funds would need to be found.[14] the substitution of generics for branded products, and so on, the health production function is dis- 3.4 Relationship with Budget Impact placed upwards (i.e. greater returns in health are obtained from healthcare inputs). As the budget The threshold-searcher model[8] can be used to

increases, the cost-effectiveness threshold should explore the relationship between budget impact and

also increase, i.e. less efficient interventions should the cost-effectiveness threshold. The threshold is the

be incorporated into the portfolio of treatments pro- inverse of the MHG per unit of expenditure of the

vided by the NHS, provided that the productivity of least efficient intervention in current use. In figure 2,

existing healthcare activities grows at a slower rate the substitution of a more for a less efficient inter-

than the budget (in figure 2, an expansion in the vention causes the MHG of the least efficient inter-

budget is represented as an extension of the line H–a vention to rise. Thus, in figure 2, after the process of

to the right – the MHG of the least efficient interven- substituting more efficient technologies for less effi-

tion that can be funded is lower the further to the cient ones, the MHG is E–b. Future interventions

right you go and thus the threshold is higher). In will have to have an MHG ≥E–b, rather than E–a, to

times of rapid expansion of the NHS budget, such as justify incorporation into the portfolio of funded

have been seen over the past 7 years, the counter- treatments. As a result, the threshold for future deci-

vailing effects of the implementation of new treat- sions decreases. The next candidate intervention

ments and increases in the budget may have made will need to be even more efficient in order to justify

the adoption of a cost-effectiveness range a (fortui- its inclusion as a funded intervention. This is the

tously) appropriate approach. Conversely, when case even if the budget impact of the substitution is

budgetary growth is less than the net budget impact neutral, i.e. when the budget impact of the new

of investment and disinvestment decisions, the cost- intervention is identical to the budget impact of the

effectiveness threshold should fall to reflect the in- displaced intervention. Thus, the cost-effectiveness

creased efficiency of the marginal intervention.

threshold is, as a matter of logic, endogenous once

These relative rates of growth of the budget and one allows for dynamic interactions, even though, in

productivity of healthcare also have implications, an overall sense, it is constrained by the budget

which do not concern us here, for discounting.[17]

determined by parliament.

To the extent that the total cost of the new inter- 3.5 Summary vention is greater than that of the procedure it re-

places, a positive recommendation requires more The budget of the NHS is set by parliament.

disinvestment until the budgetary impact of succes- NICE is charged “to appraise the clinical benefits sive substitutions is neutral and the budget con- and costs of such health care interventions as may be straint holds. This means that the cost-effectiveness notified by the Secretary of State or the National threshold for an intervention with a large budgetary Assembly for Wales … and to reach a judgement as impact should be lower than that for an intervention to whether, on balance, this intervention can be with a small impact (i.e. because the lower the recommended as a cost effective use of NHS and threshold the higher the MHG; and when the greater Personal and Social Services (PSS) resources.”[18] It the budget impact, the greater the amount of current is clear that NICE is not mandated to determine the activity that has to be displaced to fund the new budget of the NHS, and since setting a threshold technology and therefore more efficient current independently of the budget is logically equivalent

(8)

to determining the budget, NICE cannot be mandat- ness threshold implies that the full value of the ed to do that either. The appropriate approach to innovation (greater efficacy) is captured by the man- NICE’s cost-effectiveness threshold is therefore to ufacturer. As the manufacturers are typically profit see it as an equilibrating variable that promotes the maximizers, they will seek to price as close to this efficient (health maximizing) use of a fixed budget. point as possible. Strictly, what is happening is that the cost-effectiveness information is information not previously available to manufacturers about the 4. Implications of Setting the Threshold

maximum willingness to pay of the demanders and to Optimally Exhaust a Fixed Budget

makes the task of perfect price discrimination, or the use of an ‘all-or-nothing’ demand curve, more readi- 4.1 Considering Innovation

ly achievable by producers.[20]

Figure 3[19] shows the total health gain to the While it is appropriate for manufacturers to ap- NHS population under three scenarios. Consider an propriate a share of the value of innovations, it intervention costing £20 000 per patient for a health would be unwise to create a system under which gain of 2 QALYs. At this price, the ICER is below they extract it all. The public sector subsidises R&D the cost-effectiveness threshold and the net health in a number of ways, through publicly funded re- benefit of the intervention is 1 QALY per person. At search, tax incentives and research infrastructure a price of £40 000 and 2 QALYs gained, the ICER is investment. Therefore, even if society were uncon- exactly equal to the threshold, and at this point, the cerned about who benefits from innovation (NHS net benefit from the new intervention is zero, the patients or the pharmaceutical industry), it would loss of health from displaced technologies being the not be efficient to allow full appropriation of the

same as the gain. value of innovation by the manufacturer. However,

However, if the new treatment is more effective society is most certainly concerned about this distri- than existing treatments, setting the price at a level bution, and it is reasonable that at least some of the that produces an ICER equal to the cost-effective- benefits of innovation should accrue to NHS pa-

£20 000 per QALY Price = P* £40 000

Cost-effectiveness threshold

£20 000 per QALY

QALYs gained Cost

£30 000 per QALY Price > P* £60 000

3 2

£10 000 per QALY Price < P* £20 000

1

Net health benefit 1 QALY

Net health benefit –1 QALY

Fig. 3. Threshold and health gain (reproduced from Claxton et al.[19]). P* = maximum price the NHS can afford or the value of the technology.

(9)

tients. In pharmaceuticals, as in other industries possible to identify efficiency without making as- where innovation is protected, society currently per- sumptions about the relative value of additional mits monopoly rents during patent protection, but QALYs to different people. Interpersonal compari- does not allow full appropriation by, for example, sons are therefore inherent in the process of estab- facilitating perfect price discrimination. lishing efficiency.[21] An important further consider- Such concerns as this are somewhat tangential to ation relates to the wider opportunity cost of Ap- those of the Institute. However, of direct relevance praisal Committee decisions. When the threshold is for NICE is the use of innovation as an argument for being used to allocate a fixed budget, there is not recommending interventions having ICERs above one category of patient interest but two: those pa- the threshold. When the ICER is close to, or at, the tients who would receive the new treatment or some threshold value, the full value of the innovation is alternative and those patients who bear the opportu- already being paid to the manufacturer. To recom- nity cost of its provision (i.e. those whose service mend an intervention when the ICER is above the availability is reduced by virtue of the expenditure threshold is to pay more for the innovation than it is

on the new treatment).

worth (in terms of the population’s health). Promot-

We have already observed that NICE does not ing population health is consistent only with recom-

know, and probably can not know, which patients mending treatments with ICERs that are below the

bear the opportunity cost of its appraisal guidance. If threshold. It seems inappropriate for NICE to seek to

honour its obligations to promote innovation NICE recommends an intervention on equity through such a subsidy and at possible cost to NHS grounds, it necessarily has to make assumptions patients. NICE’s contribution to innovation is more about the characteristics of those patients who bear likely to be realized effectively through clarity and the opportunity cost. Specifically, in making a posi- consistency in the criteria that it uses to make its tive recommendation, it must assume that the health recommendations. The ultimate benefit is to bring gain forgone by those who bear the opportunity cost the desire of the NHS to use interventions that are no is valued less than that of those who receive the more costly than they need be into the research plans benefit.

of manufacturers, so that the market is not disrupted

Procedural justice would seem to require that the by unforeseen changes in requirements and innova-

character of the claims of the anonymous bearers of tion is of the sort that maximizes and properly

the opportunity cost be properly considered in NICE rewards industry’s contribution to the nation’s

appraisals. In particular, when claims are made by health.

advocacy and other groups about the special nature, need, etc., of the people they represent, NICE must 4.2 Equity Arguments do their best to assess the extent to which these claims carry greater weight than the claims that could be made by those bearing the opportunity cost.

The threshold represents the opportunity cost of

Given the typical pattern of NHS expenditure, the the implementation, i.e. the health gain forgone by

typical bearer of the opportunity cost is, for exam- other patients. While the threshold is critical to the

ple, likely to be elderly and in the last year of life. It determination of the most efficient (i.e. health maxi-

does not therefore appear intuitively plausible to mizing) use of NHS resources, the Appraisal Com-

suppose that the weight to be attached to benefi- mittee also considers whether there is any ground in

ciaries’ health gains must necessarily be higher than equity for weighting the health gains and losses of

that attached to the anonymous losers. Plainly this is different people differentially or for recommending

an area in which information is poor and broad technologies with relatively high ICERs on grounds

generalizations will, for some time, have to substi- of their beneficial impact on equity.[1]

tute for more specific identification of the character- While efficiency, in the sense of health max-

istics of ‘typical’ displaced health gain. The matter imization, is a major concern of NICE’s Appraisal

Committee, it is not the only one; nor, indeed, is it is ripe for research.

(10)

4.3 Monitoring and some care as they take no account of the many Recommending Disinvestment interventions that impact on quality of life rather than survival. This said, the results are consistent with a central estimate across all programme bud- The 2004 Methods Guide avoids defining an

gets around the lower limit of the current range.

explicit threshold on the grounds that the correct

The variation in thresholds between programme figure cannot be known. However, both the pre-

budgets has implications for NICE. It implies that viously suggested casuistry (building up of specific

the opportunity cost of a NICE recommendation cases) and the threshold searching model imply that

also varies depending upon where it falls, so it may it may be reasonable for NICE to utilize explicit

be efficient or inefficient dependent on local circum- thresholds that might converge over time on a ‘best

stance. The risk of NICE guidance being inefficient estimate’. A crucial part of this search process

will depend inter alia on (i) the degree to which would be the identification of activities for disin-

national resource allocation captures geographical vestment or, when there is budgetary growth, to

variation in health needs; (ii) the degree to which identify other planned investments that ought to be

local resource allocation processes reflect variations abandoned in order to fund NICE recommendations.

in health needs between patient groups; and There has been little research on either selecting

(iii) whether the technology appraisal programme is or implementing disinvestments in the NHS. NICE

focussed on those areas with the greatest potential has commissioned research from Brunel and City

for increasing the efficiency of NHS activity. Local Universities, which is yet to be published.[22] Should

commissioners will almost certainly need guidance it turn out that actual disinvestments have tended to

on how best to identify and then manage disinvest- be more cost effective than the NICE recommended

ments and postponement of planned investments interventions, there would be prima facie evidence

following NICE recommendations. It seems obvi- for supposing either that the current threshold is too

ous that the criteria they use ought not to conflict high or that NHS trusts and commissioners were

with those used by NICE (though doubtless supple- making poor decisions at their levels. Discovering

mented by further criteria).

which the case was would plainly be an important

Culyer et al.[8] suggested that NICE should ac- piece of work. However, it will not be easy to

tively make both disinvestment and investment re- discover. Local commissioners’ choices will be de-

commendations. The Institute has started to explore termined by several factors, including their total

this possibility.[24] External organizations have also budgets, cost structures, the case mixes of the popu-

started to recommend that the NHS, via NICE or lations they serve and even the ease of implementa-

other routes, should disinvest from activities not tion. As these factors vary across Primary Care

having a robust evidence base.[25] Most recently, the Trusts (PCTs), the threshold is also likely to vary by

House of Commons Select Committee recommen- PCT, and therefore whether NICE appraisal gui-

ded that NICE should appraise potential candidates dance has a positive or negative impact on the

for disinvestment, commenting that it was unaccept- efficiency of local healthcare will also vary by PCT.

able that the Institute had ignored the Committee’s Martin et al.[23] examined the actual changes in

earlier recommendation to this effect.[26] Unfortu- programme budgets and health across PCTs and

nately, the catalogue of procedures for which the estimated the average budget elasticity of health,

evidence base is poor or absent is very long and, that is, the proportionate changes in health resulting

where there is advantage to be had from disinvest- from marginal changes to programme budgets. They

ment, it is unlikely to be the case that the scale of provided empirical estimates of the cost-effective-

disinvestment required entails the entire elimination ness threshold expressed as life-years gained. They

of a procedure. So the task of specifying disinvest- reported a range from £7397 for respiratory prob-

ment guidance is by no means easy.

lems to £26 453 for diabetes mellitus (year 2005–6

values). The threshold estimates for cancer and cir- The use of a cost-effectiveness threshold is, at its culation problems were £13 931 and £8426, respec- core, about matching investment and disinvestment tively. These figures need to be interpreted with to increase the total health produce by the health

(11)

service. To date, NICE has focussed its efforts on tional source of uncertainty would be significant, investment. Knowing what in fact is disinvested given all the other major uncertainties facing indus- from can provide some insight into whether, on try (such as the high failure rates in phase III drug average, NICE appraisal guidance is improving the development), is unclear. Current pricing arrange- efficiency of the NHS. In the future, a programme of ments allow companies to amortize the cost of these disinvestment guidance, to balance the investment failures through the price of future successes, so the guidance, might give the public and the NHS greater system may even encourage unnecessarily high-risk confidence that the net benefit of the NICE Apprais- investments. Changes in the threshold could be used al Programme was positive. to signal to the pharmaceutical industry and others the changes in the efficiency in the NHS that the 5. The Changing Threshold Institute was established to promote, and allow the industry to incorporate these changes into its invest- The Chair of NICE recently observed that the

ment appraisal processes. This in turn would reduce current threshold range has been utilized for 7 years

the risk of treatments coming to market that did not and noted that the methods review process would

deliver sufficient additional health gain to justify the need to consider whether the range should change or

price consistent with an acceptable return on invest- remain the same.[27]

ment. What would also help would be the wide promulgation of the principles upon which changes 5.1 Empirical Estimation

in the threshold would be made, thus enabling future The House of Commons Health Committee changes to be anticipated.

thought that the choice of threshold was “of serious The empirical evidence of Martin et al.[23] indi- concern.”[26] The grounds for this concern were that, cated that even the lower end of the current cost-

“it is not based on empirical research and is not effectiveness range may be too high and likely to directly related to the NHS budget. It seems to be lead to less efficient treatments being implemented higher than the threshold used by PCTs for treat- at the cost of more efficient ones. The Select Com- ments not assessed by NICE.”[26] mittee report observes that the current threshold is The Institute’s response to the Health Commit- higher than the ICER used by PCTs in their commis- tee’s comments identified 17 technology appraisals sioning processes.[26] Thus, there is a prima facie that had produced costs savings and stressed that case for considering reducing the threshold. How- most of the recommendations from the clinical ever, it maybe premature to substantially change the guidelines programme – if implemented – would threshold on the basis of the current narrow range of save the NHS money.[28] It went on to highlight the studies.

range of knowledge-promotion activities it is pursu- The efficiency of NICE guidance may be pro- ing to promote efficiency in clinical practice and moted without changing the threshold. The 2004 commissioning. Whilst these are important and val- Methods Guide indicated that £20 000 is the thres- uable activities, it would seem appropriate, given the hold at which criteria other than the ICER come into mandatory nature of guidance from the Technology play. A substantial proportion of the treatments ap- Appraisal Programme, that the value of the thres- proved by NICE have been in this range. Modifying hold gives the NHS and the people it serves confi- the utilization of these ‘other factors’ in line with the dence that the opportunity cost of the programme is arguments we have set out would have the effect of less than the value of the health gain it produces. strengthening the lower bound of the current range as the effective threshold, and thus promote the 5.2 Should the Threshold Change? efficiency of future NICE guidance.

A disadvantage of using a moving/converging 6. Conclusions threshold, or one that was subject to periodic adjust-

ment, is that it would evidently introduce an addi- The incremental cost-effectiveness threshold, as tional uncertainty and provide a less secure environ- used by NICE, is a means for promoting the opti- ment for industrial innovation. Whether this addi- mum allocation of a fixed budget. It is not necessari-

(12)

The authors thank Louise Longworth, Carole Longson,

ly an expression of society’s willingness to pay for

David Barnett and members of the DSU for constructive

health. Using the threshold-searcher model de-

comments on earlier versions of this paper.

scribed by Culyer et al.[8] to explore the implications of this, we conclude the following:

References 1. It is feasible and probably desirable to operate an

1. NICE guide to the methods of health technology appraisal.

explicit single threshold rather than the current London: NICE, 2004

2. The use of thresholds: discussion. In: Towse A, Pritchard C,

range.

Devlin N, editors. Cost effectiveness thresholds: economic and

2. The threshold should be seen as a threshold at ethical issues. London: King’s Fund and Office of Health

which ‘other’ criteria are taken into account beyond Economics, 2002: 38

3. Dolan P, Shaw R, Tsuchiya A, et al. QALY maximisation and

the ICER itself.

people’s preferences: a methodological review of the litera-

3. Interventions with a large budgetary impact may ture. Health Economics 2005 Feb; 14 (2): 197-208 4. Culyer AJ. Introduction. In: Towse A, Pritchard C, Devlin N,

need to be subject to a lower threshold as they are

editors. Cost effectiveness thresholds: economic and ethical

likely to displace more than the marginal activities. issues. London: King’s Fund and Office of Health Economics, 2002: 9-15

4. Reimbursement at the threshold transfers the full

5. Rawlins MD, Culyer AJ. National Institute for Clinical Excel-

value of an innovation to the manufacturer. Positive

lence and its value judgements. BMJ 2004; 329: 224-7

decisions above the threshold on the grounds of 6. Devlin N, Parkin D. Does NICE have a cost effectiveness threshold and what other factors influence its decisions? A

innovation reduce population health.

binary choice analysis. Health Econ 2004; 13: 437-52

5. The value of the threshold should be reconsidered 7. Smith RD, Richardson J. Can we estimate the ‘social value’ of a QALY? Four core issues to resolve. Health Policy 2005; 74

regularly to ensure that it captures the impact of

(1): 77-84

changes in efficiency and budget over time.

8. Culyer AJ, McCabe CJ, Briggs AH, et al. Searching for a

6. The use of equity weights to sustain a positive threshold not setting one: the role of the National Institute for Health and Clinical Excellence. J Health Serv Res Policy 2007

recommendation when the ICER is above the thres-

12; 1: 56-8

hold requires knowledge of the equity characteris- 9. Loomes G. Valuing life years and QALYs: transferability and convertibility of values across the UK public sector. In: Towse

tics of those patients who bear the opportunity cost.

A, Pritchard C, Devlin N, editors. Cost effectiveness thresh-

Given the barriers to obtaining this knowledge and olds: economic and ethical issues. London: King’s Fund and

knowledge about the characteristics of typical bene- Office of Health Economics, 2002: 46-55

10. Williams A. What could be nicer than NICE? London: Office of

ficiaries of NHS care, caution is warranted before

Health Economics, 2004

accepting claims from special pleaders. 11. Hutton J, Maynard A. A NICE challenge for health economics.

Health Econ 2000; 8: 89-93

7. Uncertainty in the evidence base should not be

12. Devlin N. An introduction to the use of cost effectiveness issues

used to justify a positive recommendation when the in decision making: what are the issues? In: Towse A, Pritch-

ICER is above the threshold. ard C, Devlin N, editors. Cost effectiveness thresholds: eco- nomic and ethical issues. London: King’s Fund and Office of

8. The development of a programme of disinvest-

Health Economics, 2002: 16-25

ment guidance would enable the Institute and the 13. Birch S, Gafni A. Cost effectiveness/utility analyses: do current decision rules lead us to where we want to be? J Health Econ

NHS to be more confident that the net health benefit

1992; 11: 279-96

of the Technology Appraisal Programme was posi- 14. Gafni A, Birch S. Guidelines for the adoption of new technolo- gies: a prescription for uncontrolled growth in expenditures

tive.

and how to avoid the problem. CMAJ 1993; 148: 913-7 15. Birch S, Gafni A. On being NICE in the UK: guidelines for

technology appraisal for the NHS in England and Wales.

Acknowledgements Health Econ 2002; 11 (3): 185-91

16. Sendi P, Gafni A, Birch S. Opportunity costs and uncertainty in the economic evaluation of health care interventions. Health This paper was initially prepared as a briefing paper for

Econ 2002; 11: 23-31

NICE as part of the process of updating the Institute’s 2004 17. Claxton K, Culyer AJ, Sculpher M, et al. Discounting and cost Guide to the Methods of Technology Appraisal. The work effectiveness in NICE: stepping back to sort out a confusion was funded by NICE through its Decision Support Unit [editorial]. Health Econ 2006; 15 (1): 1-4

18. National Institute for Health and Clinical Excellence. Frame- (DSU), which is based at the universities of Sheffield,

work document [online]. Available from URL: http://www.

Leicester, York, Leeds and at the London School of Hygiene

nice.org.uk/NICEmedia/pdf/appendixB_framework.pdf [Ac- and Tropical Medicine.

cessed 2008 Jul 18]

C. McCabe has worked with and for many stakeholders in 19. Claxton K, Briggs A, Buxton MJ, et al. Value based pricing for the NICE appraisal process, including pharmaceutical com- NHS drugs: an opportunity not to be missed? BMJ 2008 Feb;

336: 251-4 panies, patient groups and NHS organizations.

(13)

20. Friedman M. Lectures in price theory. Chicago (IL): Aldine, line]. Available from URL: http://www.adph.org.uk/pre

1971 ss_releases/20070131.php?PHPSESSID=1jftq9htaji5qglme2i

21. Culyer AJ. The bogus conflict between efficiency and equity. 3a01nlh5 [Accessed 2008 Jul 17]

Health Econ 2006; 15: 1155-8

26. House of Commons Health Committee. National Institute for 22. Appleby J, Devlin N, Parkin D, et al. Searching for local NHS

Health and Clinical Excellence: first report of session 2007-08.

cost effectiveness thresholds: a feasibility study. NICE confer-

Vol. 1. London: The Stationary Office, 2008 Jan ence; Manchester; 2007 Dec 5-6 [online]. Available from

URL: http://www.nice2007.co.uk/ApplebyDevlin.pdf [Ac- 27. Rawlins M. The future for NICE (National Institute for Health

cessed 2008 Jul 17] and Clinical Excellence). Pharm J 2007; 278 (7452): 589

23. Martin S, Rice N, Smith P. Further evidence on the link between

28. National Institute for Health and Clinical Excellence. NICE health care spending and health outcomes in England [CHE

response to the committee’s first report of the session discussion paper 32]. York: University of York, 2007

2007-2008 HSC 550. London: The Stationary Office, 2008 Jun 24. National Institute for Health and Clinical Excellence. NICE

launches commissioning guides as part of an initiative to help NHS reduce spend on ineffective treatments [online]. Avail-

Correspondence: Professor Christopher McCabe, Academic able from URL: http://www.nice.org.uk/newsevents/infocus/

nice_launches_commissioning_guides_as_part_of_an_initiat Unit of Health Economics, Leeds Institute of Health Sci- ive_to_help_nhs_reduce_spending_on_ineffective_treatment

ences, University of Leeds, Charles Thackrah Building, s.jsp [Accessed 2008 Jul 17]

101 Clarendon Rd, Leeds, LS2 9LJ, UK.

25. Association of Directors of Public Health. Letter from the

E-mail: c.mccabe@leeds.ac.uk President: patient charges for treatment. ADPH 2007 Jan [on-

Referenzen

ÄHNLICHE DOKUMENTE

- In case you are applying for a travel grant for junior scientists: Does your supervisor still have collaborative projects with Senckenberg or the Alumnus/a respectively. (please

The preceding suggests the fol- lowing answer to the question what goal incremental confirmation is sup- posed to further: Science aims at informative truth, and one should stick

The analysis improves on earlier accounts in German descriptive linguistics in that it offers a fully compositional account of the semantic and pragmatic contribution of eigentlich in

Nelsen and Daniels‟s (2007) definition of quality is invalid according to Routio‟s (2009) criteria as the two parts of this definition ignores the requirements of the

We attempt to characterise the SARB’s inflation targeting policy by analysing the Bank’s interest rate setting behaviour before and after the adoption of inflation targeting, making

Ensuite, trois Points d’Échanges Internet (PEI) furent installés au Royaume-Uni, et d’autres pays européens firent de même. Avec l’arrivée de la concurrence dans

If buyers’ feedback-leaving behaviors are consistent with the lab findings, we will see that buyers are willing to bear the nuisance cost of leaving negative feedback (i.e., cost

Social Development Canada defines the social economy as a grass-roots entrepreneurial, not-for-profit sector, based on democratic values that seeks to enhance the social,